On Wed, Apr 25, 2012 at 3:15 PM, dan (ddp) <[email protected]> wrote:
> On Wed, Apr 25, 2012 at 9:05 AM, jfranz <[email protected]> wrote:
>> Hi guys,
>>
>> I'm pretty new to OSSEC and running in problems getting a centralized
>> configuration done.
>>
>> I'm running OSSEC-Server on a CentOS 6.2 box and OSSEC-Agent on a
>> Ubuntu 12.04 box. Everything is working fine so far (only plain
>> installation and connecting the agent to the server).
>>
>> Here comes the trouble: I changed the host-deny and firewall-drop
>> commands inside the ossec.conf of the server to location "all". As I
>> understand it, this means that i.e. if somebody is bruteforcing the
>> sshd running on the agent, the host-deny and firewall-drop active
>> responses should trigger both on the agent and the server. I my case,
>> this doesn't work. I can see the sshd bruteforce attack inside the
>> logs on the server, so the connection between agent and server works,
>> but only the agent which was under attack blocked the ip via iptables
>> and host.deny, and not the server.
>>
>
> Unfortunately I think this is the way it works. "all" actually means
> "all agents," not "all systems." I think this might be worth looking
> into...

Ok. That makes sense. Will test that by adding another machine/agent
to my testing setup.

>
>> As I understand it, the agent uses the ossec.conf and *_rules.xml
>> files from the server?!?!
>>
>
> No. Each agent uses its own ossec.conf, and possibly the agent.conf
> from the server.
>
> The rules and decoders are only handled on the server. The agent
> passes a log message to the server, which then compares it to the
> decoders/rules.
>

Ok. Got that. But how does an agent know when to use an active
response, based on a rule that is parsed on the server after the agent
send it's log?

>> I my case, the server config file has the following md5sum:
>> 1dd21647768bd23ac2a83e62adbbc0ca  ossec.conf
>>
>> The agent is using a different one:
>>
>> OSSEC HIDS agent_control. Agent information:
>>   Agent ID:   001
>>   Agent Name: Client1
>>   IP address: 10.17.0.15
>>   Status:     Active
>>
>>   Operating system:    Linux testing 3.2.0-22-generic-pae #35-Ubuntu SMP Tu..
>>   Client version:      OSSEC HIDS v2.6 / 77843f5c451af0a872a5e4733655aa1e
>>   Last keep alive:     Wed Apr 25 14:55:16 2012
>>
>>   Syscheck last started  at: Wed Apr 25 14:30:15 2012
>>   Rootcheck last started at: Wed Apr 25 14:40:43 2012
>>
>> This one is located under ../shared/agent.conf on the server:
>> 77843f5c451af0a872a5e4733655aa1e  shared/agent.conf
>>
>> The OSSEC documentation says that I can push a centralized
>> configuration from the server to the agent, by creating the file
>> ../shared/agent.conf. In my case, this file already exists and
>> contains a perl script.
>>
>
> That's wrong. I don't know why there would be a perl script there.
> Something went haywire in your installation.
>

Yepp. Did an clean install on the server and files look fine now. No
more perl script in there.

Btw: Thx for your help ;)

>> I'm stuck. So, how can I use a centralized set of rules that I created
>> in the local_rules.xml on the server and push it to the agent and how
>> can an alarm on of the agents can be triggered on all agents
>> connecting to the same server?
>>
>> I would really appreciate if anybody could help me out with this,
>> because OSSEC seems to be really nice!
>>
>> Best regards,
>> _joern_

Reply via email to